At the moment I'm completely confused whether cycleways (and pavements) next to roads should be mapped using additional ways or by tags. I'm for using additonal ways, because there are a lot of arguments. I've listed them on this page:
To start with, I want to talk about the process of mapping. This consists of two parts: First, the data is collected and assembled into an internal format (nodes, segments, ways, areas, maybe superways, all this with associated tags). Second, this data is somehow transformed to get the desired output (i.e. maps, navispeach, routing texts etc.)
In my oppinion collecting data should represent the reality as close as necessary, but as abstract as possible. This means, every way, even the tiniest one, should be mapped, because it might be of interest for someone (e.g. for hikers), whereas we are not interested in every plant or stone. It's then up to the application that does the transformation to collect the data from this internal format, that meets the needs of the desired output and to throw everything else away.
In my oppinion this standpoint is already enough to justify mapping cycleways with separate lines. But there are more arguments, being more concrete:
- When using tags for cycleways next to roads, a lot of attributes of the cycleway cannot be mapped. Among these are: Width, surface, oneway, noexit, access. A workaround might be additional tags, but I feel quite uneasy with tags like widthofleftsidecycleway=1.2m
- The cycleway may depart from a road and get back there a few ten meters later, with e.g. a house in between. Of course this could be mapped with a small extra cycleway, but this might lead to strange results when rendered on maps, because it's high probably that they won't show the cycleways next to the road, but the small departing part.
- The different types of cycleways cannot be mapped. Are pedestrians allowed? Or is it a footway where cycles are allowed, or just a cycleway? This again can be done by a workaround with tags and this again will lead to ugly tags like leftsidefootwaycycleallowed=yes
There might be some arguments against this approach and I want to state my oppinion on them:
- The data is more clear if cycleways are not mapped on their own. - This is not true. It's only the visual part that is more clear, because no parallel lines occure. In fact the tags get much more complicated.
- Mapping is much easier, because no cross connections have to be entered. - To get a high quality result, we shouldn't be lazy here. And anyway I think the discussed superways would help a lot.
- The maps do look ugly with cycleways and pavements next to a lot of streets. - This is a mistake of the rendering software, not of the part where the data is collected. Again superways will help here.
- It's hard to put so many segments and nodes in parallel, and if they aren't parallel the resulting maps do not look nice. - This is true yet, but eventually josm (or other tools) should be able to correct this and make them really parallel.
Actually, I also found one argument against it, that I cannot refute now. The street and the cycleway next to it do belong together somehow, and this fact cannot be mapped yet, but may be important. But with the introduction of superways this argument would vanish too.